The Computer Is a Moron. AI Is a Moron with a Microphone.

Why the Old Playbook for Technology Adoption Just Became Obsolete

In 1967, Peter Drucker looked at the computers flooding American offices and declared: "The computer is a moron. And the stupider the tool, the brighter the master must be." He wasn't wrong. But here's what Drucker couldn't have predicted: we'd eventually give that moron a microphone, a confident voice, and access to your customer database.

Welcome to the AI era, where 70-85% of AI initiatives fail to meet expected outcomes, and 42% of companies have abandoned the majority of their AI projects in 2025 alone, which is up from just 17% the year before.

I've watched this movie before. I've been implementing enterprise technology for 16 years, and I'm seeing companies make the same mistake they made with PCs, with CRM systems, with every major technology wave: they think adoption is something that happens TO users rather than something that happens WITH users.

Except this time, the penalty for that mistake is much steeper.

The Old Model: Deploy and Wait

When personal computers arrived in offices in the 1980s, adoption was a drag on ROI, but not a dealbreaker. A PC sitting unused on someone's desk was inefficient. The company still functioned. Email still delivered. Spreadsheets still calculated. You could wait out the late adopters.

The numbers prove it worked eventually. By 1990, only 42% of American adults used a computer in any capacity. Companies absorbed that inefficiency and called it progress.

CRM systems told the same story. Research shows that 50-55% of CRM implementations fail to deliver their intended value, with poor user adoption consistently identified as the primary culprit. Yet companies kept buying licenses, kept running training sessions, kept hoping passive exposure would eventually convert the holdouts.

That approach assumed the technology delivered baseline value regardless of user engagement. And for most enterprise software, it did.

AI Just Broke That Model

Here's what keeps me up at night: I'm watching companies implement AI the same way they implemented every other technology: deploy it, train on it once, and wait for magic to happen. Then they wonder why employees are spending their days monitoring AI systems that are constantly wrong.

The difference is fundamental. Previous technologies were tools that worked independently of user skill. AI is a partnership that only works when humans actively teach it their context, correct its mistakes, and apply their judgment to its outputs.

An AI that doesn't know your role, your processes, your customers' quirks, your industry's unwritten rules? It's just an expensive autocomplete that hallucinates 39.6% of the time.

BCG's research confirms what I'm seeing in the field: only 26% of companies have developed the capabilities to move beyond proofs of concept and generate tangible value from AI. The organizations succeeding follow an uncomfortable truth: they invest 70% of their AI resources in people and processes, and only 10% in algorithms.

The Monitoring Trap

When companies implement AI to "automate" without an adoption strategy, they create something worse than the old system. Instead of doing the work, employees now watch AI do the work wrong, fix what AI breaks, and feel less competent because they've been repositioned as checkers rather than contributors.

The research backs this up: 47% of enterprise AI users made at least one major decision based on hallucinated content in 2024. That's not an AI problem. That's a user adoption problem. It's people who were never taught how to partner with AI, now trusting outputs they shouldn't trust.

This is the worst of both worlds: all the friction of a new system with none of the benefits. And it's happening because we're treating AI like it's a fax machine when it's actually a new hire who needs onboarding.

The Path Forward

User adoption isn't the final phase of AI implementation. It's the entire implementation. Without humans actively teaching AI their domain expertise, without people understanding what AI does well and what it hallucinates, without workers who see themselves as partners rather than supervisors, the technology delivers nothing.

Drucker was right in 1967: the stupider the tool, the brighter the master must be. AI is the stupidest smart tool we've ever built. It can process everything and understand nothing without human context.

The companies winning at AI aren't the ones with the best algorithms. They're the ones who figured out that humans are the key differentiator in an AI-enabled world, and they're investing accordingly. In a world where every company uses the same AI models, the difference is HOW your people use AI.

The moron has a microphone now. Time to teach it what to say.

—

Sources:

• Drucker, Peter F. "The Manager and the Moron." McKinsey Quarterly, 1967. mckinsey.com

• AI adoption and failure rates: BCG "Where's the Value in AI?" 2024; S&P Global 2025; Fullview AI Statistics 2025. bcg.com

• PC adoption statistics: U.S. News & World Report, Pew Research Center. usnews.com

• CRM implementation failure rates: Radin Dynamics; C5 Insight research. radindynamics.com

• AI hallucination rates and enterprise decision-making: Fullview AI Statistics Roundup 2025. fullview.io